19 hours with Outlook down: the monumental failure of Microsoft 365 continues to affect companies and individual users
The fall of Microsoft 365 is already close to 19 hours. Although the company talks about a phased recovery, the service is still far from normal, both in business accounts and in personal accounts. In our tests, we have found that Outlook and Copilot do not load at allOneDrive opens but won’t show files, Office on the web doesn’t work, and Edge doesn’t sync between devices. Teams, for its part, seems to be working normally.
As we reported this morning, the problems began on Thursday at 5:17 p.m., peninsular time, and their effects have been felt throughout the world. In the United States, some companies sent their employees home early. In Australia, several schools were unable to offer online classes.
This is how the events have happened according to Microsoft’s state account
It took more than an hour and a half for Microsoft to acknowledge the drop in Exchange Online is Microsoft’s cloud email service. Shortly after, the company admitted that the bug “may affect the connectivity of Microsoft 365 services for users in any region.”
From there, the messages have continued. First, Microsoft talked about a part of the infrastructure with “a high load in several regions” and began to distribute the traffic between other centers. It then paused all deployments and restarted the affected servers.
He later clarified that the charge did not come from an increase in traffic. The problem was that several servers had unexpectedly become unavailable, overloading the rest. And why did they fall? Because they did not pass automatic health checks, that is, the periodic tests with which the system decides whether a server is working well.
Microsoft still doesn’t know what caused it
The company itself acknowledged that it had not identified “any recent changes related to the onset of the impact.” He aimed at a component responsible for processing the state of the serverstried an alternative verification method and looked into reverting some recent configuration updates. Around midnight he admitted that the measures taken had not worked and that a central process on the servers was becoming saturated.
Soon after came another setback. “Servers that we have returned to service may become unresponsive again as soon as they receive traffic,” explained Microsoft, which continued to analyze diagnostic data to find out what was causing it. Since then, the strategy has been to limit, block and redistribute traffic, with a partial recovery in the regions where it has been applied.
The latest message, posted mid-morning, talks about “a progressive improvement in several workloads.” Still, he acknowledges that “the status of the service remains variable and capacity has not yet fully returned to its baseline level.” To avoid losing held mail, Microsoft has expanded to 48 hours the period before messages expire.
